See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/23436966

The thermodynamic dissociation constants of

ambroxol, antazoline, naphazoline,
oxymetazoline and ranitidine by the regression
analysis of spectrophotometric data
ARTICLE in TALANTA MARCH 2004
Impact Factor: 3.51 DOI: 10.1016/j.talanta.2003.08.027 Source: PubMed

CITATIONS

DOWNLOADS

VIEWS

34

181

205

3 AUTHORS, INCLUDING:
Milan Meloun

Tomas Syrovy

University of Pardubice

University of Pardubice

165 PUBLICATIONS 1,063 CITATIONS

23 PUBLICATIONS 183 CITATIONS

SEE PROFILE

SEE PROFILE

Available from: Milan Meloun

Retrieved on: 17 June 2015

Talanta 62 (2004) 511522

The thermodynamic dissociation constants of ambroxol, antazoline,

naphazoline, oxymetazoline and ranitidine by the regression
analysis of spectrophotometric data
Milan Meloun a, , Tom Syrov a , Ale Vrna b
a

Department of Analytical Chemistry, University of Pardubice, Namesti Cs. Legii 565,

CZ-532 10 Pardubice, Czech Republic
b IVAX Pharmaceuticals, s.r.o. 74770 Opava, Czech Republic

Received 5 May 2003; received in revised form 7 July 2003; accepted 25 August 2003

Abstract
The mixed dissociation constants of five drug acidsambroxol, antazoline, naphazoline, oxymetazoline and ranitidineat various ionic
strengths I of range 0.01 and 1.0 and at temperatures of 25 and 37 C were determined using SQUAD(84) regression analysis of the
pH-spectrophotometric titration data. A proposed strategy of efficient experimentation in a protonation constants determination, followed
by a computational strategy for the chemical model with a protonation constants determination, is presented on the protonation equilibria
of ambroxol. The thermodynamic dissociation constant pKaT was estimated by non-linear regression of {pKa , I} data at 25 and 37 C: for
T
T
T
T
ambroxol pKa,1
= 8.05 (6) and 8.25 (4), log 21
= 11.67 (6) and 11.83 (8), for antazoline pKa,1
= 7.79 (2) and 7.83 (6), pKa,2
= 9.74 (3)
T
and 9.55 (2), for naphazoline pKTa,1 = 10.81 (1) and 10.63 (1), for oxymethazoline pKTa,1 = 10.62 (2) and 10.77 (7), pKa,2
= 12.03(3) and
T
= 1.89 (1) and 1.77 (1). Goodness-of-fit tests for various regression diagnostics enabled the reliability of
11.82 (4) and for ranitidine pKa,1
the parameter estimates to be found.
2003 Elsevier B.V. All rights reserved.
Keywords: Spectrophotometric titration; Dissociation constant; Protonation; Ambroxol; Antazoline; Naphazoline; Oxymetazoline; Ranitidine

1. Introduction
In the 1990s, the pharmaceutical industry and regulatory health care authorities adopted a new system for the
classification of drugs, the Biopharmaceutics Classification
System (BCS) [13]. BCS classifies every pharmaceutical active ingredient into one of four groups based on
two basic characteristics: solubility and permeability. The
system reflects contemporary experience in the evaluation
of the most important features of drugs which affect the
formulation of medicine preparation and the regulatory
consequences. When a poorly soluble drug is to be formulated, attention is paid mainly to an improvement of
its solubility, and thus mostly to the selection of appro Corresponding author. Tel.: +420-4660-37026;
fax: +420-4660-37068.
E-mail addresses: milan.meloun@upce.cz (M. Meloun),
tomas.syrovy@upce.cz (T. Syrovy), ales vrana@ivax-cr.com (A. Vrana).

0039-9140/$ see front matter 2003 Elsevier B.V. All rights reserved.
doi:10.1016/j.talanta.2003.08.027

priate pharmaceutical excipient(s). In more soluble drugs,

there is generally more information on their protonation
behaviour in water systems. However, the dependence of
protonation constants on ionic strength has been systematically investigated only in a few cases in the literature.
The authors decided to complete such information and
to study the protonation equilibria of five readily soluble
drugs. In three, the protonation/dissociation equilibria can
play an important role because of their site of application,
the nasal mucosa and/or eye (naphazoline, antazoline, and
oxymetazoline). At the same time, the range of osmolality of body liquids at the site of absorption (tears and
nasal secretions) is better defined and much narrower than
in cases of absorption in the gastrointestinal tract. In the
other two drugs, the protonation/dissociation behaviour on
site of absorption is generally known, and the influence of
ionic strength should not play a role, i.e. ambroxol as a
representative of basic, and ranitidin as a representative of
acidic drugs.

512

M. Meloun et al. / Talanta 62 (2004) 511522

Naphazoline and oxymetazoline belong to the group of

-sympatomimetics, agonists of -adrenergic receptors. Antazoline is a histamine H1 -receptor antagonist. They all have
vasoconstrictive effects, due to which they reduce the lumen
of capillaries, and help relieve the oedema of nasal tissue.
Antazoline, which possess an antihistaminic effect, is often used in combination with other vasoconstrictors to treat
rhinitis of allergic origin. As allergic rhinitis is accompanied
by eye irritation, the final dosage formsdropscontaining
these compounds are often designed both as nasal and eye
drops [4]. However, the pH of nasal mucosa is slightly acidic,
pH = 5.55.6, while the pH of tears, the conjunctival liquid,
is slightly basic, pH = 7.4.
Ideally, both active compounds should not be much dissociated/protonated in this pH range to achieve good absorption. On the other hand, the pH of the nasal or eye
drop formulation itself plays an important role in perception (well-being) after application on the one hand and in
the stability of the formulation on the other which may lead
to contradictory conditions.
Ranitidine is a competitive antagonist of histamine
H2 -receptors. Due to its acidic character, it is less dissociated under low pH and is thus considered a representative
of drugs well absorbed in the stomach. Nevertheless, the
dissociation constant of ranitidine is not listed in the general
literature, e.g. [5,6]. Ambroxol is a well-known secretolytic
and mucolytic drug. It is almost completely absorbed, and
due to its basic character, the sites of its absorption are
tissues with basic pH. Therefore, ambroxol is a representative of drugs which are well absorbed in either the small
intestine or rectum.
Ambroxol, chemically (2-amino-3,5-dibromo-N-[trans-4hydroxy-cyclohexyl] benzylamin,

Naphazolin, chemically 4,5-Dihydro-2-(1-naphthalenylmethyl)-1H-imidazole,

Oxymethazoline, chemically 2-(3-hydroxy-2,6-dimethyl4-tert-butylbenzyl)-2-imidazolin,

Ranitidin, chemically N,N-dimethyl-N-[5-[2-(1-methylamino-2-nitrovinylamino)-ethyl-thiomethyl]furfuryl]-amine,

This paper investigates the dissociation constants of the

five drugs: ambroxol, antazoline, naphazoline, oxymetazoline and ranitidine at various ionic strengths and at 25 and
37 C, to prove their reliability and also to estimate the thermodynamic dissociation constant pKaT at these two temperatures. The pKaT may be used for prediction of the actual dissociation constant pKa at the given value of an ionic strength.

2. Theoretical
Antazoline, chemically 4,5-Dihydro-N-phenyl-N-(phenylmethyl)-1H-imidazole-2-methanamine, 2-[(N-benzylanilino)
methyl]-2-imidazole,

Computations related to the determination of protonation

constants [710] may be performed by the regression analysis of spectra using versions of the SQUAD program family
[8,1116]. If the protonation equilibria between the anion,
L (the charges are now omitted for the sake of simplicity) of
a drug and a proton, H, are considered to form a set of variously protonated species L, LH, LH2 , LH3 , . . . etc., which
have a general formula Lq Hr in a particular chemical model
and are represented by p the number of species, (q, r)i , i =
1, . . . , p where index i labels their particular stoicheiometry, then the overall protonation constant of the protonated
species, qr , may be expressed as
qr =

[Lq Hr ]
c
= q r
[L]q [H]r
l h

(1)

M. Meloun et al. / Talanta 62 (2004) 511522

where the free concentration [L] = l, [H] = h and [Lq Hr ] =

c. For the ith solution measured at the jth wavelength, the
absorbance, Ai,j , is defined as
Ai,j =

p

j,n cn =

n=1

p

(qr,j qr lq hr )n

m
n

n=1

(Aexp,i,j Acalc,i,j )2

i=1 j=1

m
n



i=1 j=1

(Aexp,i,j

p

j,k ck )2 = minimun

for a fitness test. If, after termination of the minimization process the condition s(A) sinst (A) is met and the
R-factor is less than 1%, the hypothesis of the chemical
model is taken as the most probable one and is accepted.

(2)

where qr,j is the molar absorptivity of the Lq Hr species

with the stoichiometric coefficients q, r measured at the jth
wavelength. The absorbance Ai,j is the element of the absorbance matrix A of size (n m) being measured for n
solutions with known total concentrations of two basic components, cL and cH , at m wavelengths. Throughout this paper, it is assumed that the n m absorbance data matrix
A = C containing the n recorded spectra as rows can be
written as the product of the m p matrix of molar absorptivities and the p n concentration matrix C. Here, p is
the number of components that absorb in the chosen spectral range. The rank of the matrix A is obtained from the
equation rank(A) = min[rank(), rank(C)] min(m, p, n).
Since the rank of A is equal to the rank of  or C, whichever
is the smaller, and since rank() p and rank(C) p, then
provided that m and n are equal to or greater than p, it is only
necessary to determine the rank of matrix A, which is equivalent to the number of dominant light-absorbing components
[8,17,18]. All spectra evaluation may be performed with the
INDICES algorithm [18] in the S-Plus programming environment. Most indices methods are functions of the number of principal components PC(k)s into which the spectral
data are usually plotted against an integer index k, PC(k) =
f(k), and when the PC(k) reaches the value of the instrumental error of the spectrophotometer used, sinst (A), the corresponding index k represents the number of light-absorbing
components in a mixture, p = k . In a scree plot the value
of PC(k) decreases steeply with an increasing number PCs
as long as the PCs are significant. When k is exhausted the
indices fall off, some even displaying a minimum. At this
point p = k for all indices. The indices values at this point
can be predicted from the properties of the noise, which may
be used as a criteria to determine p [18].
The multi-component spectra analysing program
SQUAD(84) [13] may adjust qr and qr for absorption
spectra by minimising the residual-square sum function, U,
U=

513

(3)

k=1

where Ai,j represents the element of the experimental absorbance response-surface of size n m and the independent variables ck are the total concentrations of the basic
components cL and cH being adjusted in n solutions. The
calculated standard deviation of absorbance s(A) and the
Hamilton R-factor are used as the most important criteria

3. Experimental
3.1. Chemicals and solutions
All the drugs were used in the form of hydrochloride, nitrate or mesylate. Ranitidine hydrochloride was purchased
from SMS Pharmaceuticals, India, with a purity of 98.3%.
Antazoline mesylate was purchased from SIMS S.p.A., Italy,
with a purity of 99.8%. Naphazoline nitrate was purchased
from LOBA Feinchemie, Austria, with purity 99.3%. Ambroxol hydrochloride was purchased from Boehringer Ingelheim, Germany, with a purity of 99.9%. Oxymetazoline
hydrochloride was purchased from SigmaAldrich with a
purity of 99.6%. Perchloric acid, 1 M, was prepared from
conc. HClO4 (p.a., Lachema Brno) using redistilled water
and standardized against HgO and NaI with a reproducibility less than 0.20%. Sodium hydroxide, 1 M, was prepared
from pellets (p.a., Aldrich) with carbon dioxide-free redistilled water and standardized against a solution of potassium
hydrogen-phthalate using the Gran method in the MAGEC
program [11] with a reproducibility of 0.1%. Mercuric oxide, sodium iodide, and sodium perchlorate (p.a., Lachema
Brno) were not further purified. The preparation other solutions from analytical reagent-grade chemicals has been described previously [9,10]. Twice-redistilled water was used
in the preparation of solutions.
3.2. Apparatus and pH-spectrophotometric titration
procedure
The used apparatus and the pH-spectrophotometric titration procedure has been described previously [19].
3.3. Procedure for determination of the chemical model
and protonation constants
The experimental and computational schemes for the
determination of the protonation constants of the multicomponent system is taken from Meloun et al. [8] and are
described in a previous contribution [19]. When a minimization process terminates, some diagnostics are examined to
determine whether the results should be accepted: the physical meaning of parametric estimates, the physical meaning
of the species concentrations, the goodness-of-fit test and
the deconvolution of spectra.
3.4. Determination of the thermodynamic
protonation/dissociation constants
The non-linear estimation problem of the thermodynamic
dissociation constant KaT = aH+ aL /aHL , is simply a

514

M. Meloun et al. / Talanta 62 (2004) 511522

Fig. 1. Absorption spectra of the protonation equilibria of ambroxol in dependence on pH at 25 C: (a) 3D-absorbance response-surface representing
SQUAD(84) input data, (b) the 3D-overall diagram of residuals represents a response-surface indicating the quality of a goodness-of-fit.

problem of optimization in the parameter space in which the

pKa and I are known and given values while the parameters
pKa , , and C are unknown variables to be estimated [8,19].

goodness-of-fit test, cf. page 101 in Ref. [8] or may be found

in a previous paper [19].
3.6. Software used

3.5. Reliability of estimated protonation/

dissociation constants

Computations were performed by regression analysis of

UV/Vis spectra using the SQUAD(84) program [13]. The
thermodynamic dissociation constant pKT was estimated
with the non-linear regression program MINOPT in the ADSTAT statistical system (TriloByte Statistical Software Ltd.,
Pardubice) [20].

The adequacy of a proposed regression chemical model

with experimental data and the reliability of parameter estimates pKa,i found, being denoted for the sake of simplicity as bj , and ij , j = 1, . . . , m, may be examined by the
0.60
(a)

-1.5

(b)

log sk(A)
-2.5

0.30

-3.0
0.15
0.00

-3.5
280

300

[nm]

10

-4.0
100

LH

*10-3

(c)

(d)

LH
L

[%]

60

L2H

40

2
1
0

20

L2H
280

300 [nm]

7.0

7.7

pH

0
9.1

Fig. 2. Estimation of the protonation constants and molar absorptivities of ambroxol at 25 C and I = 0.006: (a) scree plot for determination of the
number of light-absorbing species in mixture k = 3 and the instrumental error of the spectrophotometer used s3 (A) = 0.25 mAU, (b) goodness-of-fit
scatter plot: s(e) and |e| bar line for each spectrum, (c) the spectra of molar absorptivities vs. wavelengths for all of the variously protonated species,
(d) distribution diagram of the relative concentrations of all of the variously protonated species.

M. Meloun et al. / Talanta 62 (2004) 511522

4. Results and discussion

4.1. Estimation of protonation/dissociation constants of
five drugs
A proposed strategy for efficient experimentation in protonation constants determination followed by spectral data
treatment is presented on the protonation equilibria of ambroxol. pH-spectrophotometric titration enables absorbance
response-surface data (Fig. 1a) to be obtained for analysis
with non-linear regression. The reliability of parameter estimates (pKs and s) may be evaluated on the basis of the
goodnes-of-fit test of residuals (Fig. 1b). The SQUAD(84)
program [13] analysis process starts with data smoothing
followed by a factor analysis using the INDICES procedure
[18]. The position of a break-point on the sk (A) = f(k)
curve in the scree plot is calculated and gives k = 3 with the
corresponding co-ordinate s3 (A) = 0.25 mAU which also
represents the instrumental error sinst (A) of the spectrophotometer used (Fig. 2a). Two protonation constants and three
molar absorptivities of ambroxol calculated for 39 wavelengths constitute 236 unknown parameters which are refined by the MR algorithm in the first run of the SQUAD
program. In the second run, the NNLS algorithm makes the
final refinement of all of the previously found parameter estimates with all molar absorptivities kept non-negative. The
reliability of the parameter estimates may be tested with the
use of SQUAD(84) diagnostics:
The first diagnostic value indicates whether all of the
parametric estimates qr and qr have physical meaning and
reach realistic values. As the standard deviations s(log qr )
of parameters log qr and s(qr ) of parameters qr are significantly smaller than their corresponding parameter estimates
(Table 1), all variously protonated species are statistically
significant at significance level = 0.05.
The physical meaning of the protonation constants, molar
absorptivities, and stoichiometric indices is examined: qr
and qr should be neither too high nor too low, and qr should
not be negative. The absolute values of s(j ), s(j ) gives information about the last U-contour of the hyperparaboloid
in neighbourhood of the pit, Umin . For well-conditioned parameters, the last U-contour is a regular ellipsoid, and the
standard deviations are reasonably low. High s values are
found with ill-conditioned parameters and a saucer-shaped
pit. The relation s(j ) F < j should be met where F is
equal to 3. The set of standard deviations of pqr for various
wavelengths, s(qr ) = f(), should have a Gaussian distribution; otherwise, erroneous estimates of qr are obtained.
Fig. 2c shows the estimated molar absorptivities of all of
the variously protonated species L , LH , L2 H ambroxol in
dependence on wavelength. Some spectra quite overlap and
such cases may cause some resolution difficulties.
The second diagnostic tests whether all of the calculated free concentrations of variously protonated species on
the distribution diagram of the relative concentration expressed in percents have physical meaning, which proved

515

Table 1
Determination of protonation constants and molar absorptivities of the
variously protonated species of ambroxol by regression analysis of the
UV/Vis absorption spectra with SQUAD(84) for n = 20 spectra measured
at m = 39 wavelengths for two basic components L and H forming p = 3
variously protonated species
Protonation constants

Partial correlation
coefficients

Lq Hr Log qr s(log qr )

L1 H1

L1 H 2

L1 H1 7.968
L2 H1 11.34

1
0.6814

0.009
0.017

Determination of the number of

light-absorbing species by factor analysis
Number of light-absorbing species, p

Residual standard deviation s3 (A) (mAU)

0.25

Goodnes-of-fit test by the statistical analysis of residuals

Hamilton R-factor (%)
Residual mean e
Mean residual |e| (mAU)
Standard deviation of residuals s(e) (mAU)
Residual skewness g1 (e)
Residual kurtosis g2 (e)

0.33
6.00 1012
0.86
1.21
0.12
2.64

The charges of the ions are omitted for the sake of simplicity.

to be the case (Fig. 2d). The calculated free concentration

of the basic components and variously protonated species
of the chemical model should show molarities down to
about 108 M. Expressed in percents, a species present at
about 1% relative concentration or less in an equilibrium
behaves as a numerical noise in regression analysis. A distribution diagram makes it easier to judge the contributions of individual species to the total concentration quickly.
Since the molar absorptivities will generally be in the range
103 105 l mol1 cm1 , species present at less than ca. 0.1%
relative concentration will affect the absorbance significantly
only if their is extremely high. The diagram shows that
overlapping protonation equilibria of H with LH and L exist.
The third diagnostic concerning the matrix of correlation
coefficients in Table 1 proves that there is an absence of
interdependence in the pair of protonation constants LH and
L2 H of ambroxol.
The fourth diagnostic concerns the goodness-of-fit
(Fig. 1b). To identify the best or true chemical model when
several are possible or proposed, and to establish whether or
not the chemical model represents the data adequately, the
residuals e should be analysed. The goodness-of-fit achieved
is easily seen by examination of the differences between
the experimental and calculated values of absorbance, ei =
Aexp,i,j Acalc,i,j . Examination of the spectra and of the
graph of the predicted absorbance response-surface through
all the experimental points should reveal whether the results
calculated are consistent and whether any gross experimental errors have been made in the measurement of the spectra.
One of the most important statistics calculated is the standard deviation of absorbance, s(A), calculated from a set of
refined parameters at the termination of the minimization

516

M. Meloun et al. / Talanta 62 (2004) 511522

Table 2
The search for a chemical equilibrium model of ambroxol using regression analysis of pH-spectrophotometric data with SQUAD(84), with the standard
deviations of the parameter estimates in the last valid digits in brackets
q, r
1,
1,
1,
2,
2,

0
1
2
1
2

Log qr

7.927 (10)

Log qr

15.773 (19)

Degree-of-fit test by the statistical analysis of residuals

R-factor (%)
1.84
3.18
s(A) (mAU)
6.53
11.27
sk (A) (mAU), p
0.4, 2
0.52, 2
e
0.0048
0.0086
s(e)
0.0065
0.0113
g1 (e)
0.87
0.17
g2 (e)
3.66
2.67
Model is
1,
1,
2,
3,
3,
4,

0
1
1
1
2
2

Rejected

7.598 (7)

14.316 (31)

Rejected

8.179 (9)

8.000 (***)

Degree-of-fit test by the statistical analysis of residuals

R-factor (%)
0.44
0.6
s(A) (mAU)
1.62
2.18
sk (A) (mAU), p
0.25, 3
0.25, 3
e
0.0001
0.0015
s(e)
0.0002
0.0022
g1 (e)
0.42
0.29
2.50
3.14
g2 (e)
Model is

Rejected

Rejected

Log qr

8.044 (9)
9.000 (1927.778)

1.13
4.13
0.25, 3
0.0028
0.0041
0.66
3.87
Rejected

Log qr

Log qr

7.968 (9)

11.340 (17)

7.952 (14)

12.000 (***)

0.33
1.21
0.25, 3
0.0009
0.0012
0.12
2.64

0.70
2.54
0.25, 3
0.0018
0.0025
0.34
3.16

Accepted

Rejected

7.900 (10)

18.500 (***)

7.936 (44)
11.902 (42)
7.000 (***)

7.061 (31)
11.736 (46)

18.500 (***)

0.84
3.06
0.25, 3
0.0022
0.0031
0.34
2.63

0.27
1.01
0.19, 4
0.0007
0.0010
0.06
2.43

0.29
1.08
0.19, 4
0.0008
0.0011
0.12
1.98

Rejected

Accepted

Rejected

The reliability of parameter estimates found is proven with goodness-of-fit statistics such as the Hamilton R-factor (%), the residual standard deviation
sk (A) (mAU) and the standard deviation of absorbance after termination of the regression prooess, s(A) (mAU); (***) means that the estimate of the
standard deviation is too large.

process. It is usually compared with the standard deviation of absorbance calculated by the INDICES program
[18], sk (A), and if s(A) sk (A), or s(A) sinst (A), the
instrumental error of the spectrophotometer used, the fit
is considered to be statistically acceptable (Table 2). This
proves that the s3 (A) value is equal to 0.25 mAU and is quite
close to the standard deviation of absorbance when the minimization process terminates, s(A) = 1.21 mAU. Although
this statistical analysis of residuals [22] gives the most
rigorous test of the degree-of-fit, realistic empirical limits
must be used. For example, when sinst (A) s(A) 0.003,
the goodness-of-fit is still taken as acceptable, whereas
s(A) > 0.010 indicates that a good fit has not been obtained. Alternatively, the statistical measures of residuals e
can be calculated: the residual mean (known as the bias) e
should be a value close to zero; the mean residual |e| and
the residual standard deviation s(e) should be close to the
absorbance standard deviation sinst (A); the skewness g1 (e)
should be close to zero for a symmetric distribution; the kurtosis g2 (e) should be close to 3 for a Gaussian distribution;
a Hamilton R-factor of relative fit, expressed as a percent-

age (R 100%), of <0.5% is taken as an excellent fit, but

>2% is a poor one. The statistical measures of all residuals
e proves that the minimum of the eliptic hyperparaboloid U
is reached (Table 2): the residual mean e = 6.00 1012
proves that there is no bias or systematic error in spectra
fitting. The mean residual |e| = 0.86 mAU and the residual standard deviation s(e) = 1.21 mAU have sufficiently
low values. The skewness g1 (e) = 0.12 is quite close to
zero and proves a symmetric distribution of the residuals
set, while the kurtosis g2 (e) = 2.64 is close to 3 proving
a Gaussian distribution. The Hamilton R-factor of relative
fitness is 0.33%, proving an excellent achieved fitness, and
therefore the parameter estimates may be considered as
suitably reliable.
The fifth diagnostic, the spectra deconvolution in Fig. 3,
shows the deconvolution of the experimental spectrum into
spectra for the individual, variously protonated species.
Spectrum deconvolution seems to be quite useful tool in the
proposal of a strategy for efficient experimentation. Such
a spectrum provides sufficient information for a regression
analysis which monitors at least two species in equilibrium

M. Meloun et al. / Talanta 62 (2004) 511522

517

0.60
LH + L + L 2H

(a)

LH + L + L 2 H

LH

(b)

LH
0.30

0.30
0.15

L 2H

L 2H

L
0.00

0.15

280

[nm]

300

0.45 LH + L + L 2 H

280

0.00

[nm]

300

LH + L + L 2 H

(c)

(d)

A
LH
L

0.15

LH
L 2H

0.00

A
0.30

0.30

0.45

280

300

0.15

L 2H
[nm]

280

300

[nm]

0.00

Fig. 3. Deconvolution of the experimental spectrum of ambroxol into spectra for the individual variously protonated species in solution for pH equal to:
(a) 8.4, (b) 8.0, (c) 7.6, and (d) 6.6.

where none of them is a minor species. The minor species

has a relative concentration in a distribution diagram of less
than 5% of the total concentration of the basic component
cL . When, on the other hand, only one species is prevalent
in solution, the spectrum yields quite poor information in a
regression analysis, while the parameter estimate is rather
unsure, and is definitely not reliable enough.

In searching for the best chemical model of protonation

equilibria, 10 various hypotheses of the stoichiometric indices q and r of Lq Hr acid were tested in order to find that
which best represented the data (Table 2). The criteria of
resolution used for the hypotheses were: (1) a failure of the
minimisation process in a divergency or a cyclisation; (2)
an examination of the physical meaning of the estimated

Table 3
Dependence of the mixed dissociation constants of ambroxol on ionic strength using regression analysis of pH-spectrophotometric data with SQUAD(84),
with the standard deviations of the parameter estimates in the last valid digits in brackets
Determined chemical model at 25 C contains L, LH, L2 H
Ionic strength
log 11
log 21
Goodness-of-fit test
R-factor (%)
sk (A) (mAU)
s(A) (mAU)

0.006
7.968 (9)
11.640 (17)
0.33
0.25
1.21

0.019
8.035 (15)
11.670 (38)
0.41
0.29
1.57

Determined chemical model at 37 C contains L, LH, L2 H

Ionic strength
0.006
0.019
log 11
log 21
Goodness-of-fit test
R-factor (%)
sk (A) (mAU)
s(A) (mAU)

8.194 (11)
11.860 (33)
0.56
0.2
1.87

8.184 (14)
11.917 (36)
0.54
0.21
1.98

0.033
8.050 (9)
11.509 (21)

0.046
7.933 (9)
11.480 (18)

0.059
8.029 (11)
11.626 (27)

0.072
7.908 (15)
11.546 (27)

0.38
0.25
1.55

0.42
0.23
1.7

0.34
0.23
1.39

0.5
0.23
2.03

0.033

0.046

0.059

0.072

8.160 (9)
11.850 (26)
0.47
0.3
1.92

8.192 (17)
12.018 (43)
0.69
0.2
2.73

8.190 (22)
11.830 (58)
0.51
0.23
2.02

8.200 (19)
12.004 (48)
0.54
0.24
2.10

The reliability of parameter estimates found is proven with goodness-of-fit statistics such as the Hamilton R-factor (%), the residual standard deviation

sk (A) (mAU) and the standard deviation of absorbance after terminatiori of the regression process, s(A) (mAU) at 25 C (upper part) and 37 C (lower part).

518

M. Meloun et al. / Talanta 62 (2004) 511522

Table 4
Dependence of the mixed dissociation constants of antazoline on ionic strength using regression analysis of pH-spectrophotometric data with SQUAD(84)
with the standard deviations of the parameter estimates in the last valid digits in brackets
Determined chemical model contains L, LH, LH2 at 25 C
Ionic strength
pKa,1

0.010
9.778 (25)

0.010
9.721 (26)

0.070
9.516 (36)

0.089
9.535 (29)

0.127
9.478 (30)

0.271
9.559 (20)

0.402
9.459 (23)

0.794
9.275 (11)

0.925
9.297 (9)

0.25
0.2
0.81

0.34
0.27
1.22

0.33
0.22
1.13

0.29
0.16
1.03

0.25
0.23
0.90

0.35
0.21
1.25

0.24
0.15
0.91

0.25
0.27
0.90

Determined chemical model contains L, LH, LH2 at 25 C

Ionic strength
0.010
0.030
0.071
7.694 (34)
7.626 (53)
7.546 (52)
pKa ,2

0.141
7.530 (46)

0.271
7.524 (41)

0.402
7.449 (50)

0.663
7.460 (47)

Goodness-of-fit test
R-factor (%)
0.25
sk (A) (mAU)
0.2
s(A) (mAU)
0.81

0.28
0.31
0.99

0.25
0.23
0.90

0.35
0.21
1.25

0.35
0.43
1.26

Determined chemical model contains L, LH, LH2 at 37 C

Ionic strength
0.010
0.170
0.206
pKa,1
9.532 (35)
9.315 (28)
9.298 (24)

0.411
9.206 (20)

0.467
9.211 (16)

0.491
9.206 (18)

0.571
9.188 (24)

0.598
9.178 (14)

Goodness-of-fit test
R-factor (%)
0.54
sk (A) (mAU)
0.17
s(A) (mAU)
1.60

0.45
0.23
1.46

0.33
0.21
1.14

0.54
0.28
1.65

0.34
0.25
1.31

0.45
0.16
1.55

Determined chemical model contains L, LH, LH2 at 37 C

Ionic strength
0.010
0.170
0.337
pKa,2
7.679 (43)
7.361 (49)
7.051 (45)

0.467
7.000 (53)

0.571
7.003 (55)

0.598
7.058 (34)

Goodness-of-fit test
R-factor (%)
0.54
sk (A) (mAU)
0.17
s(A) (mAU)
1.60

0.34
0.21
1.14

0.52
0.25
1.61

0.45
0.16
1.55

Goodness-of-fit test
R-actor (%)
0.19
sk (A) (mAU)
0.13
s(A) (mAU)
0.64

0.40
0.29
1.38

0.34
0.27
1.22

0.45
0.19
1.43

0.54
0.19
1.67

0.30
0.19
1.03

0.44
0.18
1.46

The reliability of parameter estimates found is proven with goodness-of-fit statistics such as the Hamilton R-factor (%), the residual standard deviation
sk (A) (mAU) and the standard deviation of absorbance after termination of the regression process, s(A) (mAU) at 25 C (upper part) and 37 C (lower part).

Table 5
Dependence of the mixed dissociation constants of naphazoline on ionic strength using regression analysis of pH-spectrophotometric data with SQUAD(84),
with the standard deviations of the parameter estimates in the last valid digits in brackets
Determined chemical model contains L, LH at 25 C
Ionic strength
pKa,1
Goodness-of-fit test
R-factor (%)
sk (A) (mAU)
s(A) (mAU)

0.009
10.767 (5)
0.42
0.28
1.40

0.026
10.767 (6)
0.30
0.31
1.05

Determined chemical model contains L, LH at 37 C

Ionic strength
0.009
0.021
pK0 ,1
10.582 (4)
10.569 (30)
Goodness-of-fit test
R-factor (%)
sk (A) (mAU)
s(A) (mAU)

0.37
0.33
1.22

0.37
0.25
1.40

0.038
10.761 (7)
0.28
0.17
1.00
0.034
10.553 (8)
0.37
0.33
1.31

0.050
10.736 (6)
0.25
0.24
0.88
0.046
10.551 (8)
0.35
0.33
1.24

0.062
10.757 (9)
0.37
0.42
1.28
0.058
10.540 (7)
0.27
0.24
0.97

0.074
10.735 (9)
0.36
0.43
1.25
0.074
10.544 (8)
0.34
0.30
1.20

0.086
10.725 (5)
0.26
0.19
0.88
0.086
10.522 (5)
0.34
0.26
1.14

The reliability of parameter estimates found is proven with goodness-of-fit statistics such as the Hamilton R-factor (%), the residual standard deviation
sk (A) (mAU) and the standard deviation of absorbance after termination of the regression process, s(A) (mAU) at 25 C (upper part) and 37 C (lower part).

M. Meloun et al. / Talanta 62 (2004) 511522

519

0.60

-1.4
(a)

(b)

logsk(A)

0.30

-2.8

0.15
-3.5
225

250 [nm] 300

12 LH2

(c)

10
100

*10

LH

(d)

LH

LH2

-3

L 60

40

20
225

250 [nm] 300

7.2

8.0

pH

9.6

Fig. 4. Estimation of protonation constants and molar absorptivities of antazolin at 25 C and ionic strength I = 0.070: (a) scree plot for determination of
the number of light-absorbing species in mixture k = 3 and s3 (A) = 0.27 mAU, (b) goodness-of-fit scatter plot: s(e) and |e| bar line for each spectrum,
(c) the spectra of molar absorptivities vs. wavelengths for all of the variously protonated species, (d) distribution diagram of the relative concentrations
of all of the variously protonated species.

parameters if they were both realistic and positive; and (3)

the residuals should be randomly distributed about the predicted regression spectrum and systematic departures from
randomness were taken to indicate that either the chemical
model or parameter estimates were unsatisfactory.

Using the experimental and evaluation strategy, the protonation equilibria of ambroxol (Table 3 and Figs. 13),
antazoline (Table 4 and Fig. 4), naphazoline (Table 5 and
Fig. 5), oxymethazoline (Table 6 and Fig. 6) and ranitidine (Table 7 and Fig. 7) were also examined. To test the

0.45

(b) -2.4

(a)

logsk(A)

-3.0
0.30
-3.6

0.15

-4.2
260

270 [nm] 290

10
100

7
(c)

*10-3

LH

(d)

LH

60

40

20
260

270 [nm] 290

10

pH

12

Fig. 5. Estimation of the protonation constants and molar absorptivities of naphazoline at 25 C and ionic strength I = 0.009: (a) scree plot for
determination of the number of light-absorbing species in mixture k = 2 and s2 (A) = 0.28 mAU, (b) goodness-of-fit scatter diagram plots s(e) and |e|
bar line for each spectrum, (c) the spectra of molar absorptivities vs. wavelengths for all of the variously protonated species, (d) distribution diagram of
the relative concentrations of all of the variously protonated species.

520

M. Meloun et al. / Talanta 62 (2004) 511522

Table 6
Dependence of the mixed dissociation constants of oxymetazoline on ionic strength using regression analysis of pH-spectrophotometric data with
SQUAD(84), with the standard deviations of the parameter estimates in the last valid digits in brackets
Determined chemical model contains L, LH, LH2 at 25 C
Ionic strength
pKa,2
pKa,1
Goodness-of-fit test
R-factor (%)
sk (A) (mAU)
s(A) (mAU)

0.009
11.998 (9)
10.623 (16)

0.023
11.984 (7)
10.595 (13)

0.19
0.13
0.84

0.079
11.806 (6)
10.408 (14)

0.24
0.22
0.89

0.25
0.15
0.89

Determined chemical model contains L, LH, LH2 at 37 C

Ionic strength
0.012
0.066
pKa,2
11.784 (29)
11.740 (23)
10.526 (76)
10.457 (77)
pKa,1
Goodness-of-fit test
R-factor (%)
sk (A) (mAU)
s(A) (mAU)

0.48
0.11
1.56

0.135
11.787 (6)
10.549 (13)
0.24
0.17
0.72

0.108
11.564 (16)
10.133 (77)

0.65
0.12
1.2

0.176
11.637 (37)
10.076 (30)

0.48
0.12
1.13

0.51
0.15
1.31

0.228
11.759 (9)
10.577 (17)
0.12
0.14
0.4
0.262
11.665 (32)
9.799 (83)
0.38
0.16
1.07

The reliability of parameter estimates found is proven with goodness-of-fit statistics such as the Hamilton R-factor (%), the residual standard deviation
sk (A) (mAU) and the standard deviation of absorbance after termination of the regression process, s(A) (mAU) at 25 C (upper part) and 37 C (lower part).

reliability of protonation/dissociation constants at different

ionic strength the goodness-of-fit test with the use of statistical analysis of the residuals was applied, and results
appear in Tables 37. For all five drugs studied the most
efficient tools, such as the Hamilton R-factor, the mean
residual and the standard deviation of residuals were applied: as the R-factor in all cases reaches a value of less
then 0.5% an excellent fitness and reliable parameter estimates are indicated. The standard deviation of absorbance

0.60

s(A) after termination of the minimization process is always

better than 2 mAU, and the proposal of a good chemical
model and reliable parameter estimates are proven.
Another problem concerns small differences of molar
absorptivities in the variously protonated species within a
spectrum (Figs. 2c, 4c, 5c, 6c and 7c). It may happen that
non-linear regression fails when small differences of absorbance are of the same magnitude as the instrumental
noise, sinst (A).

-1.6

(a)

(b)

log sk(A)

0.30
-3.2
0.15
288

296 [nm] 312 0

L

(c)

*10

-3

LH2

LH

-4.0
10
(d) 80

L %

40

LH

20

LH2
288

296 [nm] 312 9.8

10.5

pH

11.9

Fig. 6. Estimation of the protonation constants and molar absorptivities of oxymethazoline at 25 C and I = 0.023: (a) scree plot for determination of the
number of light-absorbing species in mixture k = 3 and s3 (A) = 0.22 mAU, (b) goodness-of-fit scatter plot: s(e) and |e| bar line for each spectrum, (c)
the spectra of molar absorptivities vs. wavelengths for all of the variously protonated species, (d) the distribution diagram of the relative concentrations
of all of the variously protonated species.

M. Meloun et al. / Talanta 62 (2004) 511522

521

Table 7
Dependence of the mixed dissociation constants of ranitidine on ionic strength using regression analysis of pH-spectrophotometric data with SQUAD(84),
with the standard deviations of the parameter estimates in the last valid digits in brackets
Determined chemical model contains L, LH at 25 C
Ionic strength
pKa,1

0.009
1.961 (1)

0.021
1.933 (2)

0.079
2.020 (1)

0.116
2.089 (1)

0.195
2.115 (1)

0.292
2.160 (1)

0.456
2.231 (1)

Goodness-of-fit test
R-factor (%)
sk (A) (mAU)
s(A) (mAU)

0.31
0.12
1.40

0.12
0.11
0.51

0.2
0.14
0.77

0.21
0.13
0.81

0.13
0.15
0.50

0.12
0.13
0.46

0.16
0.13
0.57

Determined chemical model contains L, LH at 37 C

Ionic strength
0.015
0.057
pKa,1
1.828 (1)
1.889 (1)

0.112
2.030 (1)

0.232
2.002 (1)

0.298
2.044 (1)

0.379
2.064 (1)

0.486
2.114 (1)

Goodness-of-fit test
R-factor (%)
sk (A) (mAU)
s(A) (mAU)

0.32
0.13
1.14

0.21
0.16
1.26

0.24
0.12
0.82

0.41
0.13
1.33

0.12
0.11
0.21

0.43
0.12
1.59

0.17
0.16
0.74

The reliability of parameter estimates found is proven with goodness-of-fit statistics such as the Hamilton R-factor (%), the residual standard deviation
sk (A) (mAU) and the standard deviation of absorbance after termination of the regression process, s(A) (mAU) at 25 C (upper part) and 37 C (lower part).

Surprisingly, at the pH of the typical absorption conditions

of ambroxol, pH = 7.7, almost one-third of the drug exists
in the form of dimer. As the molecular weight of ambroxol
is 378.11, its dimer already exceeds the (approximate) limit
for molecular weight above which the size of molecule does
play a role in absorption (ca. Mr 500).
Naphazoline and antazoline are weak bases which are
less protonated and thus, in theory, should be better absorbed at the pH of tears (pH 7.4) than of nasal mucosa
(pH 5.5). On the other hand, from the stability point of
view, the higher is the pH of the liquid preparation, the more

susceptible are both drugs to hydrolysis. This is particularly

true in the case of antazolin, which undergoes hydrolysis to N-benzylanilinoacetylethylene diamine in aqueous
formulations. As the absorption of antazoline like most
H1 antagonists is generally good, antazoline should be
formulated in a preparation whose pH compromises the best
stability with the best and most comfortable perception after
application. In general, for weak bases, the best stability can
be expected at pH equal to or less than the half value of pK.
In the case of antazoline, it would be a pH ranging from 3.9
to 4.8, which is too low with respect to the site of application
-2.5

0.60
A
0.45

(a)

(b) log s (A)

k

-3.0

0.30

-3.5

0.15

-4.0

285
16

300

[nm]

330

8
100

(c)

*10-3

LH

(d)

[%]
60

40
4
0
285

20

LH
300

[nm]

330 1.6

2.4 pH

3.2

Fig. 7. Estimation of the protonation constants and molar absorptivities of ranitidin at 25 C and I = 0.009: (a) the scree plot for determination of the
number of light-absorbing species in mixture k = 2 and s2 (A) = 0.12 mAU, (b) the goodness-of-fit scatter plot: s(e) and |e| bar line for each spectrum,
(c) the spectra of molar absorptivities vs. wavelengths for all of the variously protonated species, (d) the distribution diagram of the relative concentrations
of all of the variously protonated species.

522

M. Meloun et al. / Talanta 62 (2004) 511522

Table 8
Thermodynamic dissociation constants for ambroxol, antazoline, naphazoline, oxymetazoline and ranitidine at two selected temperatures
Value at 25 C

Value at 37 C

8.05 (6)
11.67 (6)

8.25 (4)
11.83 (8)

7.79 (2)
9.74 (3)

7.83 (6)
9.55 (2)

Ambroxol

T
pKa,1
T
log 21

Antazoline

T
pKa,1
T
pKa,2

Naphazoline

pKaT

10.81 (1)

10.63 (1)

Oxymethazoline

T
pKa,1
T
pKa,2

10.62 (2)
12.03 (3)

10.77 (7)
11.82 (4)

Ranitidine

T
pKa,1

1.89 (1)

1.77 (1)

considered. Nevertheless, from the graph of occurrence of

different protonated forms of antazoline at different pH, it
can be stated that an anthazoline formulation should have
pH under 6.46.8, where double-protonated species of antazoline dominate. The dissociation constant of antazoline
was found electrochemically [21] pKa = 10.10 at 25 C.
The unknown parameter pKaT was estimated by applying
a DebyeHckel equation to the data in Tables 37 according to the regression criterion, Table 8 shows point estimates
of the thermodynamic dissociation constants of five drugs
at two temperatures. Because of the narrow range of ionic
strengths the ion-size parameter and the salting-out coefficient C could not be estimated.
5. Conclusions
When drugs are poorly soluble then instead of a
potentiometric determination of dissociation constants,
pH-spectrophotometric titration may be used with the
non-linear regression of the absorbance response-surface
data. The reliability of the dissociation constants of five drug
acids (i.e. ambroxol, antazoline, naphazoline, oxymetazoline and ranitidine) may be proven with goodness-of-fit
tests of the absorption spectra measured at various pH.
Goodness-of-fit tests for various regression diagnostics
enabled the reliability of the parameter estimates to be
determined.
Acknowledgements
The financial support of the Internal Grant Agency of the
Czech Ministry of Health (Grant No. NB/7391-3) and of

the Ministry of Education (Grant No. MSM253100002) is

gratefully acknowledged.

References
[1] G.L. Amidon, H. Lennernas, V.P. Shah, Pharm. Res. 12 (1995) 413.
[2] US Department of Health and Human Service, FDA Center for Drug
Evaluation and Research (CDER) Draft Guidance: Waiver of In Vivo
Bioavailability and Bioequivalence Studies for Immediate Release
Solid Dosage Forms Containing Certain Active Moieties/Active Ingredients Based on a Biopharmaceutic Drug Classification System.
CDERGUID\2062 DFT.WPD, 6 January 1999.
[3] The Merck Index, An Encyclopedia of Chemicals, Drugs and Biologicals, 13th ed., Whitehouse, 1996, pp. 15282001.
[4] D. Gherghel, Optometry Today 14 (June) (2002) 2428.
[5] M. Hohnjec, J. Kuftinec, M. Malnar, M. kreblin, F. Kajfe, A.
Nagl, N. Blaevic, Anal. Profiles Drug Subst. 15 (1986) 533561.
[6] S.C. Ruckmick, D.F. Marsh, S.T. Duong, J. Pharm. Sci. 84 (1995)
502507.
[7] F.R. Hartley, C. Burgess, R.M. Alcock, Solution Equilibria, Ellis
Horwood, Chichester, 1980.
[8] M. Meloun, J. Havel, E. Hgfeldt, Computation of Solution Equilibria, Ellis Horwood, Chichester, 1988, p. 220.
[9] M. Meloun, M. Pluharov, Anal. Chim. Acta 416 (2000) 5568.

[10] M. Meloun, P. Cernohorsk,

Talanta 52 (2000) 931945.
[11] D.J. Leggett (Ed.), Computational Methods for the Determination of
Formation Constants, Plenum Press, New York, 1985, pp. 99157;
D.J. Leggett (Ed.), Computational Methods for the Determination
of Formation Constants, Plenum Press, New York, 1985, pp. 291
353.
[12] J. Havel, M. Meloun, in: D.J. Leggett (Ed.), Computational Methods
for the Determination of Formation Constants, Plenum Press, New
York, 1985, 19 pp.;
J. Havel, M. Meloun, in: D.J. Leggett (Ed.), Computational Methods
for the Determination of Formation Constants, Plenum Press, New
York, 1985, 221 pp.
[13] M. Meloun, M. Javurek, J. Havel, Talanta 33 (1986) 513524.
[14] D.J. Leggett, W.A.E. McBryde, Anal. Chem. 47 (1975) 1065.
[15] D.J. Leggett, Anal. Chem. 49 (1977) 276.
[16] D.J. Leggett, S.L. Kelly, L.R. Shine, Y.T. Wu, D. Chang, K.M.
Kadish, Talanta 30 (1983) 579.
[17] J.J. Kankare, Anal. Chem. 42 (1970) 1322.

[18] M. Meloun, J. Capek,

P. Mikk, R.G. Brereton, Anal. Chim. Acta.
423 (2000) 5168.
[19] M. Meloun, D. Burkonov, T. Syrov, A. Vrna, Anal. Chim. Acta
486 (2003) 125141.
[20] ADSTAT 1.25, 2.0, 3.0 (Windows 95), TriloByte Statistical Software
Ltd., Pardubice, Czech Republic.
[21] A.C. Andrews, J.K. Romary, J. Chem. Soc. (1964) 405.
[22] M. Meloun, J. Militk, M. Forina, Chemometrics for Analytical
Chemistry, vol. 2, PC-Aided Regression and Related Methods, Ellis
Horwood, Chichester, 1994, p. 289;
M. Meloun, J. Militk, M. Forina, Chemometrics for Analytical
Chemistry, PC-Aided Statistical Data Analysis, vol. 1, Ellis Horwood,
Chichester, 1992.